Human fibroblast growth factor-1 (FGF-1) is a potent human mitogen for a variety of cell types including vascular endothelial cells, and can stimulate such cells to develop neovasculature capable of relieving ischemia. For this reason, FGF-1 is an angiogenic factor with potential applicability in “angiogenic therapy.”1-3 FGF-1 belongs to the β-trefoil superfold.4-5 This molecular architecture is characterized by a pseudo-3-fold axis of structural symmetry, with the repeating motif being a pair of anti-parallel β-strands, known as the “β-trefoil fold.” These repeating structural motifs comprise a total of 12 β-strands that associate to form a six-stranded β-barrel capped at one end by three β-hairpins (forming the “β-trefoil” superfold; FIG. 1). Residue positions 13-17 (using the 140 amino acid form of FGF-1 numbering scheme) of the N-termini (β-strand 1), and 131-135 of the C-termini (β-strand 12), hydrogen bond with each other as a pair of anti-parallel β-strands within the six-stranded β-barrel, closely juxtaposing the two termini. When considering the three-fold symmetry of the overall architecture, the N- and C-termini are structurally related to two β-hairpin turns at positions 49-52 and 90-93 (FIG. 1). Thus, the termini in the native structure represent a break in the mainchain continuity that forms the β-barrel.
An analysis of correlated anisotropic thermal factors in a 1.10 Å atom-resolution x-ray structure of FGF-1, has identified the N- and C-termini β-strands (β-strands 1 and 12, respectively) as demarcating a boundary of domain motion within FGF-1.6 In the solution NMR structure of FGF-1 the interaction between β-strands 1 and 12 is only consistently defined through residue position 133 in β-strand 12, and the remaining positions 134-135 appear largely disordered.7 Thus, these data are consistent with the N- and C-termini β-strand interaction representing a region of structural weakness in FGF-1 and therefore potentially contributing to the unfolding process. Of additional interest, quenched-flow hydrogen exchange experiments with FGF-1 have shown that the hydrogen bonds linking the N- and C-termini anti-parallel β strands appear to be the first detectable event in the folding of FGF-1, and may provide a structural framework for subsequent folding events.8 Thus, in addition to unfolding, the interaction of the N- and C-termini β-strands may be a key contributor to the folding of FGF-1.
In an effort to study the contribution of the N- and C-termini β-strands to the stability and folding of FGF-1, Cys mutations were introduced into each β-strand with the intention of linking them through a disulfide bond. In this case, stability and folding studies under oxidizing and reducing conditions might elucidate the contribution of the N- and C-termini β-sheet formation to these processes. Two potential sites for such pair-wise mutations were identified at positions 12 and 134, and 13 and 135, respectively. These two pair wise Cys mutants were constructed and initial stability studies were performed under oxidizing conditions. The Cys 13/Cys 135 mutant exhibited a substantial decrease in stability and was not studied further. In contrast, the Cys 12/Cys 134 mutant exhibited a substantial increase in stability, suggesting that the introduced disulfide bond had stabilized the structure. However, repeating the stability studies under reducing conditions resulted in a further gain in stability. Therefore, the increase in stability for the Cys 12/Cys 134 mutant was due to the substitution of Lys 12 and/or Pro 134 by Cys and not to disulfide bond formation. As a consequence of this initial result, additional Thr and Val point mutations were constructed at positions 12 and 134 to probe the nature of the stability increase afforded by the Cys mutations. The results of these studies show that the Cys residue, in each case, is not essential and similar or greater increases in stability can be realized with Val mutations.
Isothermal equilibrium denaturation, folding and unfolding kinetics, and x-ray structural studies have been utilized in characterizing the effects of Cys, Thr and Val mutations at positions 12 and 134 in FGF-1. The results show that mutations at both positions 12 and 134 contribute to increased stability, with position 12 mutations primarily increasing the rate of folding, and position 134 mutations primarily decreasing the rate of unfolding. The combined position 12 and 134 Val mutation also exhibits a 30-fold increase in mitogenic potency and may find useful application as a “second generation” form of FGF-1 in angiogenic therapy.
Val mutations at the symmetry-related positions of residues 12 and 134 were also studied and in one case (position 95) provide a substantial additional increase in stability. A combined mutation, involving Val mutations at five positions, and introducing a three-fold symmetric constraint at two positions within the FGF-1 structure, results in an increase in stability that doubles the original value of the AG of unfolding. This combined mutation is, however, functionally inactive. The results provide additional support to our hypothesis9 that a symmetric primary structure within a symmetric superfold is a solution to, and not a constraint upon, the protein folding problem.
Furthermore, the results also support the “function/stability trade-off” hypothesis10-14, and lead us to propose that one property of the β-trefoil superfold (and presumably all the protein superfolds) is the capacity for profound thermal stability, permitting a wide-range of adaptive radiation in function.